Paper Title
Text to Speech in Kannada with Emotions

Abstract
The goal of the “Text to Speech in Kannada with Emotions” project is to translate Kannada text into speech with a human-like tone characterized by emotions. With the help of Text classification, Emotion detection and other NLP tasks, this project will fulfill the goal. The project helps to make reading easier for the visually impaired and senior citizens. With the added emotions, the final output seeks to sound more human. The project’s goal is to identify emotions from text and convert text in Kannada language to speech in Kannada language. The main objective of the project is to assist the visually impaired and those with learning disabilities like dyslexia. The project also aims at making online content more accessible to such people. The project has two objectives. The first task is to classify the emotion from Kannada text along with the class tag from the dataset created by us to perform natural language processing tasks. The emotions we try to classify text into are happiness, sadness, anger and neutral. The Kannada text post pre-processing and emotion detection will be output as text with an appropriate emotion tag. The second task is to convert this Kannada text to Kannada speech by trying to introduce the identified emotion in speech generated. Additionally, we alter the speech elements to make it sound real. Keywords - Multinomial Logistic Regression, Decision Tree Classifier, PyDub, gTTS, Tokenization, Feature Extraction, Kannada Text, Emotion Classification and Prediction, Text to Speech